AstroAI Workshop 2025
Marc Huertas-Company
Representation learning and simulation based inference for astrophysics (galaxy formation)
Presenter: Marc Huertas-Company (Instituto de Astrofísica de Canarias, University of Paris)
Title: Representation learning and simulation based inference for astrophysics (galaxy formation)
Date/Time: Friday, July 11th, 9:30 AM - 11:00 AM
Abstract: In this lecture, I will first review the concept of representation learning and its applications in astrophysics, tracing its evolution from basic approaches to recent advances in multimodal foundation models (FMs). I will then discuss several applications—with an emphasis on survey astrophysics—ranging from multimodal comparisons of observations and simulations to the transfer of learned representations to downstream tasks. Finally, I will highlight aspects of simulation-based inference (SBI) and explore how FMs could help alleviate some of its limitations, such as model drift.
Biography: Marc Huertas-Company is a staff research scientist and group leader at the Instituto de Astrofísica de Canarias (Spain) and Associate Professor - on leave - at the University of Paris and the Paris Observatory (France).
Marc’s research focuses on trying to leverage the latest AI advances to learn a bit more about how galaxies form and evolve. This includes working with amazing observing facilities such as Euclid and JWST.